Sub-ambient cooling is conventionally accomplished through gas/liquid vapor compression based refrigeration cycles using Freon type refrigerants to implement the heat transfers. Such refrigeration systems are used extensively for cooling human residences, foods, and vehicles. Sub-ambient cooling is also often used with major electronic systems such as mainframe computers. Though vapor compression cooling can be very efficient, it does require significant moving hardware, including at a minimum, a compressor, a condenser, an evaporator, and related coolant transfer plumbing. As a result of the complexity and associated high cost, vapor compression cooling has not found material acceptance in small cooling applications, for example personal computers.
The fact that CMOS logic can operate materially faster as the temperature decreases has been well known for at least ten years. For example, if CMOS logic devices are operated at -50.degree. C., the performance is improved by 50 percent over room ambient temperature operation. Liquid nitrogen operating temperatures, in the range of -196.degree. C., have shown 200 percent performance improvements. Similar benefits have shown to accrue for integrated circuit wiring, where metal wiring resistances decrease by a factor of 2 for integrated circuits operated at -50.degree. C. in comparison to room ambient operation. This improvement rivals the recent technological breakthrough of using copper wiring in integrated circuits to reduce interconnect resistance and thereby effectively increase the operating frequencies attainable. Thus, sub-ambient operation of integrated circuit logic devices, such as field effect transistors, as well as the interconnect wiring can materially improve the integrated circuit performance, leaving the question of how to accomplish such cooling in the confines of an ever decreasing size and materially shrinking cost environment.
Thermoelectric cooling is one alternative that has found some usage given the compact size of the prevalently used Peltier devices. Peltier device thermoelectric cooling is also very reliable in that the cooling is totally solid state. The key negative aspect of thermoelectric cooling is the inefficiency, wherein a Peltier device cooling system efficiency is commonly only in the range of 20 percent for a relatively nominal temperature drop between the cold sink and the ambient. For example to cool at the rate of one watt at a sub-ambient temperature of 0.degree. C. the Peltier cooling system must be powered with 5 watts. As the amount of heat to be transferred increases, the total power to be dissipated into the ambient mandates large convection devices and high output power supply circuits. Therefore, Peltier device thermoelectric cooling has not been considered a broadly applicable technology for improving integrated circuit performance.
To understand how the present invention improves thermoelectric cooling efficiency it is necessary to understand why Peltier device thermoelectric cooling is inefficient. A Peltier device is fabricated from semiconductor material such as bismuth telluride or lead telluride. Though new materials are now being evaluated in various universities, they have yet to reach fruition. The commonly used Peltier materials exhibit very high electrical conductivity and relatively low thermal conductivity, in contrast to normal metals which have both high electrical and thermal conductivity. In operation the Peltier devices transport electrons from a cold sink, at temperature T.sub.cold, to a hot sink, at temperature T.sub.hot, in response to an electric field formed across the Peltier device. However, there are other mechanisms affecting Peltier device efficiency, which mechanisms degrade the net transport of the heat energy from the cold sink to the hot sink.
FIG. 1 schematically depicts a conventional Peltier type thermoelectric element (TE) 1 with DC power supply 2 created the electric field across TE 1 while at a load current 3. The desired heat transfer is from cold sink 4, at temperature T.sub.cold, to hot sink 6, at temperature T.sub.hot. As indicated in the equation of FIG. 1, the net heat energy transported is composed of three elements, the first representing the Peltier effect (thermoelectric) contribution, the second defining negative Joule heating effects, and the third defining negative conductivity effects. The thermoelectric component is composed of the Seebeck coefficient, the temperature of operation (T.sub.cold) and the current being applied. The Joule heating component reflects that roughly half the Joule heating goes to the cold sink and remainder to the hot sink. Lastly, the negative component attributable to thermal conduction represents the heat flow through the Peltier device, as defined by the thermal conductivity of the Peltier device, from the hot sink to the cold sink. See equation (1). EQU q=.alpha.T.sub.cold I-1/2I.sup.2 R-K.DELTA.T (1)
Since the thermoelectric component of the heat transport increases in direct proportion to the current, while the Joule heating increases in proportion to the square of the current, and the thermal conduction is in direct proportion to the hot sink to cold sink temperature difference, the equation clearly reflects how quickly the Peltier device becomes inefficient.
Equation (2) defines a coefficient of performance for the Peltier device. The coefficient is the ratio of the net heat energy transported at low temperature to the power consumed in the Peltier device. For a typical bismuth telluride material Peltier device, the coefficient of performance is less than 0.3. ##EQU1## Note that the numerator of equation (2) represents the net cooling capability of the Peltier device. The denominator of equation (2) represents the total energy provided by external power source 2. The individual elements of the numerator were described earlier. The first term in the denominator is the total Joule heating, while the second term is the heat energy transport work done by the Peltier device in moving energy from the T.sub.cold sink to the T.sub.hot sink. Based upon this relationship, the maximum coefficient of performance possible in the configuration of FIG. 1 is given by equation (3). ##EQU2##
The parameter .gamma. can be expressed in terms of the Seebeck coefficient .alpha., electrical conductivity .sigma. and thermal conductivity .lambda. as set forth in equation (4). ##EQU3##
Note that the first factor in equation (3) is the Carnot efficiency, which is the maximum efficiency possible for any heat pump operating between two temperature sinks T.sub.cold and T.sub.hot. The second factor represents the non-ideal thermoelectric cooling, which can also be characterized by a figure of merit ZT. Note that .pi..sub.max .fwdarw.(T.sub.cold /.DELTA.T) as .gamma..fwdarw..infin..
To date it has been very difficult to develop a thermoelectric material which yields high values of ZT. The prevalent materials for thermoelectric coolers have been bismuth telluride (Bi.sub.2 Te.sub.3) and lead tellurium (PbTe). These materials have ZT values of approximately 0.3 at room temperature. Recent work at universities has shown that ZT values approaching 1 may be possible in lead tellurium quantum wells and multilattices. However, even with these materials, the thermoelectric cooling is not competitive with mechanical vapor compression cooling systems.
Another constraint of Peltier device cooling is the limited temperature excursion below ambient attainable. That limitation arises from the fact that temperature span is constrained by efficiency, a parameter which degrades quickly as the temperature differential increases. The maximum temperature differential possible T.sub.max is given by equation (5). EQU .DELTA.T.sub.max =1/2ZT.sup.2.sub.cold (5)
For bismuth telluride having a ZT of approximately 0.3, T.sub.max is 45.degree. K at 300.degree. K.
Thus, there are a number of very fundamental constraints on efficiency and differential temperature that limit the use of conventional thermoelectric elements for sub-ambient cooling applications.